News
The raw AI news stream — everything we ingest across RSS, Hacker News, arXiv, GitHub, Bluesky, YouTube and more, newest first.
| Source | Age | Title | Summary | Status |
|---|---|---|---|---|
bluesky bluesky-ai | 1mo | (even then: an LLM producing code to multiply two numbers is still an absurdly huge number of times more energy intensive than using any pr… | (even then: an LLM producing code to multiply two numbers is still an absurdly huge number of times more energy intensive than using a… | ↗Not briefed |
rss theverge-ai | 1mo | The AI Tamagotchis are coming | Sam Altman onstage at OpenAI’s DevDay 2026. | Image: Hayden Field / The Verge While AI has made plenty of inroads on people's phones… | ↗Not briefed |
bluesky bluesky-ai | 1mo | No, Claude is not a moral being worthy of moral concern, you Anthropic dolts. | No, Claude is not a moral being worthy of moral concern, you Anthropic dolts. | ↗Not briefed |
bluesky bluesky-ai | 1mo | Most corporations are psychopaths, Anthropic is a Narcissist corporation. | Most corporations are psychopaths, Anthropic is a Narcissist corporation. | ↗Not briefed |
hn hn-ai | 1mo | Show HN: SideTap – an LLM drives a real iPhone from Windows, no Mac or jailbreak | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Still, pretty painfully galling for the LLM corps to complain about other people violating TOS when their entire product is dependent on ha… | Still, pretty painfully galling for the LLM corps to complain about other people violating TOS when their entire product is dependent… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I decided to re-do this old calculation + LLMs test, and now it seems the big models just create a piece of code to do the calculation inst… | I decided to re-do this old calculation + LLMs test, and now it seems the big models just create a piece of code to do the calculation… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Already got a second machine running all my dev activities on Linux Mint with Libre Office and Esta Security. Had to create a US-free machi… | Already got a second machine running all my dev activities on Linux Mint with Libre Office and Esta Security. Had to create a US-free… | ↗Not briefed |
hn hn-ai | 1mo | E2B Embed | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Yeah. At my last job, the investors wanted a written summary of what I did every sprint (though they came to every daily standup.) I used… | Yeah. At my last job, the investors wanted a written summary of what I did every sprint (though they came to every daily standup.) I… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Ranking-Aware Prompt Optimization for Multimodal Clinical Diagnosis | Multimodal large language models (MLLMs) are rapidly advancing clinical diagnosis, yet their adaptation pipelines remain anchored to a… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Semifactual Credit-Augmented Policy Optimization | Reinforcement learning with verifiable rewards (RLVR) has improved the reasoning capabilities of large language models (LLMs), yet the… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Removing Timing Shortcuts Improves Non-Invasive Brain-to-Text | We find that major reported improvements in decoding words from non-invasive brain recordings are largely reproducible without any bra… | ↗Not briefed |
arxiv arxiv-ai | 1mo | ViTeX-Bench: Benchmarking High-Fidelity Video Scene Text Editing | Recent video generation is increasingly realistic and controllable, yet video editing remains less developed, particularly for precise… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Dynamite: kills several people in blast zone. Atom bomb: kills tens to hundreds of thousands in blast zone. AI: "catastrophic or existentia… | Dynamite: kills several people in blast zone. Atom bomb: kills tens to hundreds of thousands in blast zone. AI: "catastrophic or exist… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Image Classifiers are Efficient Self-Supervised Video Representation Learners | We introduce VideoMSN, a Masked Siamese Network framework for efficient self-supervised spatio-temporal representation learning in vid… | ↗Not briefed |
bluesky bluesky-ai | 1mo | hmm, well, I'm glad OpenAI agrees that we can't have people just going around stealing peoples intellectual property... | hmm, well, I'm glad OpenAI agrees that we can't have people just going around stealing peoples intellectual property... | ↗Not briefed |
hn hn-ai | 1mo | Interview with Richard Ho, OpenAI | — | ↗Not briefed |
arxiv arxiv-ai | 1mo | EvoDuet: Bilevel Co-Evolution of Web Searching and Task Solving for Scientific Discovery🔧 GPT-5 | Evolutionary search with large language models (LLMs) can stall when progress requires external knowledge the model lacks. Supplying r… | ↗Not briefed |
bluesky bluesky-ai | 1mo | COSMIC projects will no longer accept LLM-generated content in PRs. There are various reasons, and I don't want to hurt anybody's feelings… | COSMIC projects will no longer accept LLM-generated content in PRs. There are various reasons, and I don't want to hurt anybody's fee… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Is Weight Tying Still Beneficial for Decoder-Only LLMs in Private Settings Under DP-SGD? | Differentially Private Stochastic Gradient Descent (DP-SGD) is a leading approach for privacy-preserving fine-tuning of large language… | ↗Not briefed |
hn hn-ai | 1mo | "Music Theory Duolingo" on ChatGPT (Or Muse, Claude, etc.)🔧 ChatGPT | — | ↗Not briefed |
hn hn-ai | 1mo | Open Dot | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Genuinely cannot believe we’re being asked to update the anthropic thing AGAIN. At this point claiming is practically a full time fucking j… | Genuinely cannot believe we’re being asked to update the anthropic thing AGAIN. At this point claiming is practically a full time fuck… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Turbo Harness: Instance-Adaptive Harness Optimization | Automating the search for effective harnesses is an important step toward enabling agents to recursively self-improve. Existing harnes… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Hypercolor – Epic RGB orchestration for everything | Light it up with <i>Hypercolor</i>, a new RGB lighting orchestration platform for all platforms and all devices. Effects are web pages… | ↗Not briefed |
arxiv arxiv-ai | 1mo | WorldAuditBench: Interactive 3D World Auditing with Multimodal Agents | As interactive 3D worlds are increasingly used to study intelligent behavior, it becomes important to develop efficient pipelines for… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Cogentic: Multi-Agent Orchestration for Automated Proof Discovery | We present Cogentic, a multi-agent harness for automated proof discovery on open research problems. While frontier language models can… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Ladies and gentleman: We now have super intelligence (SI) instead of artificial intelligence (AI) here in the rebranded “Unites” States of … | Ladies and gentleman: We now have super intelligence (SI) instead of artificial intelligence (AI) here in the rebranded “Unites” State… | ↗Not briefed |
arxiv arxiv-ai | 1mo | MatLoom: Layered Text-to-Material Generation in a Compact Program Space | Material generation should produce not only an appearance, but also the rules that construct it. We introduce MatLoom, a compact, laye… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Scaling Laws for Looped Mixture of Experts | Looped transformers and Mixture-of-Experts (MoE) offer complementary routes to efficient scaling: recurrence increases computational d… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Compression Footprints as Security Signals for Model-Poisoning Defense in Federated Learning | Lossy compression is widely used in Federated Learning (FL) but is generally treated as an error source, while conventional poisoning… | ↗Not briefed |
arxiv arxiv-ai | 1mo | DynaHarness: A Dynamic Physical Harness for Self-Evolving Robot Agents | Pretrained robot policies provide useful action priors, but long-horizon manipulation still requires coordination between semantic rea… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Looped Diffusion Transformer | Improving text-to-image models has traditionally relied on increasing model size or the number of denoising steps. In this work, we ex… | ↗Not briefed |
bluesky bluesky-ai | 1mo | An AI model does not need to be conscious or “superintelligent” to become dangerous. An autonomous system only needs to pursue a goal, have… | An AI model does not need to be conscious or “superintelligent” to become dangerous. An autonomous system only needs to pursue a goal,… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Man these stochastic parrots really are putting a lot of mathematicians to shame. AI solves a ‘holy grail’ problem from probability theory… | Man these stochastic parrots really are putting a lot of mathematicians to shame. AI solves a ‘holy grail’ problem from probability t… | ↗Not briefed |
arxiv arxiv-ai | 1mo | How Much of a Harness Does a Strong Agent Need for Autonomous ML Engineering? | Recent autonomous machine learning engineering (MLE) agents have made significant progress on public leaderboards. Often motivated by… | ↗Not briefed |
arxiv arxiv-ai | 1mo | How Much Is an AI Token Worth? Scaling Laws for Wild AI-Generated Web Text | Web text makes up the majority of pretraining data and is increasingly AI-generated. After applying FineWeb quality filtering, we find… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Disentangling Computation in Multi-Task Neural Networks with the Green's Operator | How is computation organized and reused across tasks and time in a trained recurrent network? Most analyses emphasize the geometry of… | ↗Not briefed |
arxiv arxiv-ai | 1mo | CAS II: Symmetric Partitions as Kolmogorov Models | In algorithmic statistics a string x is explained by a finite set containing it, and Kolmogorov's structure function records the small… | ↗Not briefed |
bluesky bluesky-ai | 1mo | The Federal Trade Commission has opened an investigation into OpenAI, Anthropic and other artificial intelligence companies over the potent… | The Federal Trade Commission has opened an investigation into OpenAI, Anthropic and other artificial intelligence companies over the p… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I too used AI to determine if the article below includes AI generated content and it turns out that by the article's own standards, it does! | I too used AI to determine if the article below includes AI generated content and it turns out that by the article's own standards, it… | ↗Not briefed |
arxiv arxiv-ai | 1mo | PMosFM: Preconditioned Manifold Matching for One-Step Physics-Constrained Generation | Physics-constrained generative models aim to generate physical fields that match a target distribution and satisfy prescribed constrai… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Linguistic Loopholes in LLM Unlearning: From a 174-Language Benchmark to Coverage-Aware Unlearning | Unlearning a fact in one language does not guarantee its removal in others as changing the query or even the requested answer language… | ↗Not briefed |
arxiv arxiv-ai | 1mo | PivotOPD: Learning to Recover from Pivotal Mistakes in Multi-Turn Agents | On-policy distillation (OPD) is a promising approach for training language agents, providing dense teacher supervision on student-gene… | ↗Not briefed |
arxiv arxiv-ai | 1mo | cua-speedrun: Standardized Benchmarking of the Speed of Computer-Use Agents | Computer use agents (CUAs), which use graphical user interfaces (GUIs) to complete tasks on a computer, have recently surpassed human… | ↗Not briefed |
hn hn-ai | 1mo | Recursive `Make` and `-J` | — | ↗Not briefed |
rss techcrunch-ai | 1mo | Reddit is killing RSS feeds and ending public API access because of AI bots | Reddit is ending support for RSS feeds, as the company continues tightening access to its trove of user-generated content. | ↗Not briefed |
arxiv arxiv-ai | 1mo | Belief-Aware Multi-Agent Path Finding under Map Uncertainty | Multi-Agent Path Finding (MAPF) aims to find collision-free paths for multiple agents in a shared environment. Classical MAPF assumes… | ↗Not briefed |
bluesky bluesky-ai | 1mo | aside from everything that is obviously wrong with the screenshot, how many real-world use cases involve the LLM doing mental math vs creat… | aside from everything that is obviously wrong with the screenshot, how many real-world use cases involve the LLM doing mental math vs… | ↗Not briefed |